Fig. 1

The workflow of reconstructing HGT strains from pair-ends shotgun reads. Firstly, we map paired-end shotgun reads against reference genomes using Burrows Wheeler Aligner (BWA). Then, we select junction reads, whose two sides are mapped to two different references, from the set of mapped reads. Thirdly, by treating junction reads as points on a two-dimensional plane, we apply Density-Based Spatial Clustering of Applications with Noise (DBSCAN) to find candidate HGT breakpoints. Fourthly, we utilize split reads to get the exact positions of HGT breakpoints. Fifthly, according to the detected breakpoints, we could split references into segments, which are linked by junction reads. The coverage of each segment is calculated according to the number of mapped reads on it. Sixthly, we utilize the linked segments to construct a connected graph. By inserting dummy edges, we make the graph fully connected. Seventhly, we balance the coverage of each segment. Finally, we traverse the graph to find local strains. Each local strain should start from the first segment of one receptor and end with the last segment of the same receptor